Re: Core dumped

2010-05-28 Thread Michael McCandless
Also, are you indexing largish documents? Lucene must fully index the doc, and then flush, so for such large docs it can easily use more than the 50 MB buffer you allotted. There were some recent memory leak fixes for such large documents, as well, that you might be hitting. Which Lucene version

Re: Core dumped

2010-05-27 Thread Saurabh Agarwal
I will check it out!! Saurabh Agarwal On Thu, May 27, 2010 at 11:13 PM, Erick Erickson wrote: > The larger your RAMbufferSize, the more memory you consume FWIW. > > OK, then, does it always OOM on the same document? Are you trying to index > any particularly large documents? > > Erick > > On Thu

Re: Core dumped

2010-05-27 Thread Erick Erickson
The larger your RAMbufferSize, the more memory you consume FWIW. OK, then, does it always OOM on the same document? Are you trying to index any particularly large documents? Erick On Thu, May 27, 2010 at 1:28 PM, Saurabh Agarwal wrote: > RAMBufferSize id 50 Mb, i tried with 200 too > the index

Re: Core dumped

2010-05-27 Thread Saurabh Agarwal
RAMBufferSize id 50 Mb, i tried with 200 too the index is unoptimized MergeFactor is Default 10 and I have not changed it MaxBuffered Docs is also default Saurabh Agarwal On Thu, May 27, 2010 at 10:31 PM, Erick Erickson wrote: > What have you set various indexwriter properties to? Particularly

Re: Core dumped

2010-05-27 Thread Erick Erickson
What have you set various indexwriter properties to? Particularly things like merge factor, max buffered docs and ram buffer size. The first thing I'd look at is MergeFactor. From the JavaDocs: Determines how often segment indices are merged by addDocument(). With smaller values, less RAM is used

Core dumped

2010-05-27 Thread Saurabh Agarwal
Hi, when I am running Lucene on a 512 MB system. I am getting the following error Exception in thread "main" java.lang.OutOfMemoryError: Java heap space at org.apache.lucene.index.DocumentsWriter$ByteBlockAllocator.getByteBlock(DocumentsWriter.java:1206) and sometimes An unexpected error h